Previous studies have suggested that the attenuation of Canada lynx (Lynx canadensis) cyclic dynamics with decreasing latitude may be the consequence of a reduced specialization on the lynx's primary prey, snowshoe hares (Lepus americanus). However, intraguild competitive interactions remain largely unexplored in situations where the temporal dynamics of food resources is pronounced, and lynx populations in the south of their distribution may be negatively affected by interspecific competition with other carnivores. In this paper, we used spectral analysis of fur harvest data collected at the state (US) and province (Canada) level to explore the spatial gradient of cyclic dynamics in lynx. Although some patterns were consistent with the 'diet specialization' hypothesis, we found that temporal variance of cycling propensity peaked at mid-latitudes, where transient, non-cyclic periods, coexisted with regular 10-year cycles. In these mid-latitude zones, non-cyclic periods did not coincide with loss of snowshoe hare cycling as demonstrated by historical records, and were not more frequent in recent decades as could be expected under a 'climatic forcing' scenario. Instead, we show that non-cyclic periods tended to coincide with periods of high coyote (Canis latrans) abundance and periods when coyotes apparently tracked snowshoe hare abundance as suggested by significant 10-year cycles lagging one or two years behind hare peaks. We used landscape-scale (trapline) fur harvest returns from five provinces in Canada to further probe the importance of interspecific competition in Canada lynx population dynamics. Accounting for coyote distribution and abundance did not bring additional explanatory and predictive power to models based solely on environmental and autecological predictors, suggesting that competition with coyote is not a force driving population abundance and cyclicity among lynx. We discuss the possible factors behind the apparent lack of consistency across spatial scales and recommend that further studies examine species interactions at a smaller (local) scale.Abbreviations: GLM-Generalized Linear Model, IAD-N-inflation-adjusted + detrended harvest number, IA-N-inflation-adjusted harvest number, HPD-Highest Posterior Density, RLRT-Restricted Loglikelihood Ratio Test.